BASIC CONCEPTS IN CATEGORY THEORY EMILY RIEHL 1. Basic Definitions Definition 1.1. A category C consists of: (i) A collection of objects ob C denoted by A; B; C; : : : (ii) A collection of morphisms mor C denoted by f; g; h; : : : (iii) A rule assigning to each f 2 mor C two objects dom f and cod f, its domain and codomain. We f write f : dom f ! cod f or dom f / cod f . (iv) For each pair (f; g) of morphisms with cod f = dom g we have a composite morphism gf : dom f ! cod g subject to the axiom h(gf) = (hg)f whenever gf and hg are defined. (v) For each object A we have an identity morphism 1A : A ! A, subject to the axioms 1Bf = f = f1A for all f : A ! B. Definition 1.2. We say a category C is small if it has only a set of objects and a set of morphisms. We say a category C is locally small if for any two objects A; B of C the collection of all morphisms A ! B in C is a set. We denote this set by C(A; B). Given a category C, its opposite category Cop has the same objects but with the domain and codomain operations interchanged (and thus composition is reversed). Hence, when we prove a theorem in category theory, we simultaneously prove a \dual theorem," obtained by replacing all the categories involved with their opposites. Definition 1.3. Let C and D be categories. A functor F : C!D consists of (i) a mapping A 7! FA : ob C! ob D (ii) a mapping f 7! F f : mor C! mor D such that dom F f = F (dom f), cod F f = F (cod f), F (1A) = 1FA, and F (gf) = (F g)(F f) whenever gf is defined in C. Definition 1.4. Let C; D be two categories and F; G : C / D two functors. A natural transformation α : F ! G consists of a mapping A 7! αA ob C! mor D such that αA : FA ! GA for all A and α FA A / GA F f Gf αB FB / GB commutes for any f : A ! B in C. Note that, given another functor H and another transformation β : G ! H we can form the composite βα defined by (βα)A = βAαA. The composition is associative and has identities so we have a category [C; D] of functors C!D and natural transformations between them. Definition 1.5. We say a morphism f : A ! B is a monomorphism if fg = fh ) g = h for all g; h : C ! A. Dually, f is an epimorphism if kf = `f ) k = ` for all k; ` : B ! C. This handout was produced for the Warm-up Program (inexplicably abbreviated as \WOMP") for incoming graduate students at the University of Chicago. It is based off lectures by Peter Johnstone in his Part III course in Category theory at Cambridge in 2006. Some of the original LaTeX is due to Inna Zakharevich. 1 2. Limits and Colimits Definition 2.1. Let J be a category (almost always small or finite). By a diagram of shape J we mean a functor D : J !C. The objects D(j) for j 2 ob J are called vertices of D and the morphisms D(α) for α 2 mor J are called edges of D. For any object A of C and any J we have a constant diagram ∆A of shape J all of whose vertices are A and all of whose edges are 1A. By a cone over D : J !C with summit A we mean a natural transformation λ : ∆A ! D. Equivalently, this is a family (λj : A ! D(j) j j 2 ob J) of morphisms (the legs of the cone) such that A commutes for any α : j ! j0 in J. Note that ∆ is a functor C! [J; C]. { DD λj { D λj0 {{ DD {{ DD }{{ D(α) D" D(j) / D(j0) Definition 2.2. A limit of a diagram D : J !C is a cone (A; λ) that is universal among cones over D. That is, given another cone (B; δ) over D, there is a unique morphism f : B ! A such that δj = λjf for each j 2 J. The dual notion is a colimit. A cone under D : J !C with summit A is a natural transformation λ : D ! ∆A.A colimit of D is a cone (A; λ) under D that is universal among cones under D. That is, given another cone (B; δ) under D, there is a unique morphism f : A ! B such that δj = fλj for each j 2 J. Lemma 2.3. Limits and colimits, when they exist, are unique up to unique isomorphism. That is, given two limits (A; λ) and (A0; λ0) of a diagram D, there exists unique isomorphisms f : A ! A0 and g : A0 ! A that commute with the legs of the cones. Definition 2.4. A limit of a diagram D : J !C is a cone (A; λ) that is universal among cones over D. That is, given another cone (B; δ) over D, there is a unique morphism f : B ! A such that δj = λjf for each j 2 J. Definition 2.5. We say that C has limits of shape J if every diagram D : J !C has a limit. This is equivalent to saying that the functor ∆ : C! [J; C] has a right adjoint (see Section 4). Examples 2.6. (a) If J = ; then [J; C] has a unique object and the category of cones over it is isomorphic to C. So a limit for this diagram is a terminal object ∗ of C, i.e. one such that there is a unique morphism A ! ∗ for all A. A colimit for it is called an initial object. (b) If J is a discrete category, a diagram of shape J is just a family of objects of C, and a cone over Q it is a family of morphisms (λj : A ! D(j) j j 2 ob J). A limit for it is a product j2ob J D(j). P Similarly a colimit for this diagram is a coproduct j2ob J D(j). f (c) Let J be the finite category · / · (so a diagram of shape J is a parallel pair A / B ). A / g / h k cone over such a digram is of the form ACo / B such that fh = k = gh, or equivalently a morphism h : C ! A satisfying fh = gh. A limit for the diagram is called an equalizer for (f; g) (and a colimit for it is a coequalizer for (f; g)). (d) Let J be the finite category ··o / · . Then a diagram of shape J is a pair of morphisms g f B / CAo with common codomain. A cone over this has the form i D / A @ @@ ` k @@ @@ @ BC satisfying fh = ` = gk or equivalently a completion of the diagram to a commutative square. A terminal such completion is called a pullback for the pair (f; g). If C has products and equalizers 2 fπ e 1 then it has pullbacks: form the product A × B and then the equalizer E / A × B // C . gπ2 Then π e E 1 / A is a limit for A π2e f g B B / C g f A colimit of shape J op (i.e. of a diagram CAo / B ) is called a pushout of (f; g). Theorem 2.7. (i) If C has equalizers and all small (resp. all finite) products, then C has all small (resp. all finite) limits. (ii) If C has pullbacks and a terminal object, then C has all finite limits. Definition 2.8. Let F : C!D be a functor, J a (small) category. • We say F preserves limits of shape J if, given D : J !C and a limit cone (λj : L ! D(j) j j 2 ob J) the cone (F λj : FL ! FD(j) j j 2 ob J) is a limit cone for FD in D. • We say F reflects limits of shape J if given D : J !C and a cone (λj : L ! D(j) j j 2 ob J) such that (F λj : FL ! FD(j) j j 2 ob C) is a limit for FD, then the original cone was a limit for D. • We say that F creates limits of shape J if, given D : J !C and a limit (µj : M ! FD(j) j j 2 ob J) for FD, there exists a cone (λj : L ! D(j) j j 2 ob J) over D mapping to a limit for FD, and any such cone is a limit in C. (Note that if we require M to be in the image of F then category equivalences might not create limits, as M may not be in the image of the equivalence. This definition says that if there is a limit for FD in D then there is a limit for D in C that maps to a limit of FD in D.) Corollary 2.9. Let F : C!D be a functor. In any version of the above theorem 2.7 we may replace \C has" by either \C has and F preserves" or \D has and F creates." 3. Equivalences of Categories Definition 3.1. Let C and D be two categories. By an equivalence between C and D we mean a pair of functors F : C!D, G : D!C together with natural isomorphisms α : 1C ! GF and β : FG ! 1D.